Home / Featured /  Big Data and Loan Servicing

 Big Data and Loan Servicing

By Patrick O’Connor

Big Data is the latest event in the accelerating pace of human change. Humans started farming between 6,000 to 9,000 B.C., or about 8,000 to 11,000 years ago. In Western history, the Industrial Revolution started about 1800. The Information Revolution became mainstream in 1976 with the first Apple computer. Big Data became a mainstream concept around 2010. The time from farming to the Industrial Revolution was about 7,000 years, to computers another 180 years, and to Big Data only 30 years. The next step after Big Data will be the development of systems to analyze the data and use it for decision making. What comes next is uncertain and unknown, but it will be here quickly.

Loan Servicing in the Good Old Days

Few in the workforce remember loan servicing before Lewis Ranieri created securitization for single-family loans in 1977. Mr. Ranieri is recognized in both Liars’ Poker and in The Big Short. Before he started purchasing and securitizing thousands and hundreds of thousands of home loans, the institution that originated them serviced them. They were balance sheet loans made by banks and savings and loans. Incredible in today’s world, before 1977, the person who made the loan often handled problems with servicing. But there were few problems. The last payable to default on was typically the home mortgage.

 Big Data Described

Big Data has many definitions. Mine is simple:  Big Data combines a variety of data sources that facilitate understanding, insights, and decision making. Less important than having gargantuan-size data tables is having a variety of well-selected datasets whose confluence provides clarity in understanding complex and intractable problems. The U.S. census is a huge dataset but does not qualify as  Big Data. The addresses, loan amounts, and payment history for a million loans would not alone qualify as  Big Data. However, start adding in factors such as the value of collateral, loan-to-value, date the loan was originated, the company that originated the loan, the loan officer, and quality of loan documentation (routine or NINJA or stated income), and you can start, for example, to quickly discern patterns and start to estimate which loans are at risk.

Loan Servicing is Now Much More Complicated

Today, there are still loans that are made and serviced by the same person; however, these are jumbo loans for affluent borrowers who have other relationships with the bank. Securitization has created a new industry known as loan servicing. If the loan officer who made the loan is not available to speak to the borrower, who do they call? There are two factors about loan servicing that are indisputable: 1) Loan servicing can be lucrative, but the fees are very competitive, and 2) if a loan has been securitized, it is difficult, if not impossible, to get authority from the bond holders to modify the loan. (Once a loan is securitized, the tranches—differently levels of risk such as AAA, AA, A, BBB, BB, etc.—have been sold to different institutional investors who have disparate interests.)

However, those are the easy changes. Congress has passed legislation, including the Dodd-Frank Act, for which the regulatory agencies still have not written enabling legislation, which is inconsistent and virtually assures that every mortgage servicer is subject to inadvertently violating the law. Just a few of the requirements of Dodd-Frank are:

  • Each servicer of any federally related mortgage loan shall notify the borrower in writing of any assignment, sale, or transfer of the servicing of the loan to any other person.
  • If any servicer of a federally related mortgage loan receives a qualified written request from the borrower (or an agent of the borrower) for information relating to the servicing of such loan, the servicer shall provide a written response acknowledging receipt of the correspondence within five days (excluding legal public holidays, Saturdays, and Sundays) unless the action requested is taken within such period.
  • Damages and costs – Whoever fails to comply with any provision of this section shall be liable to the borrower for each such failure in the following amounts…
  • Class actions – In the case of a class action, an amount equal to the sum of…
  • Fail to respond within 10 business days to a request from a borrower to provide the identity, address, and other relevant contact information about the owner or assignee of the loan exposes the loan servicer to fines and penalties.

Lawsuits and Consent Agreements

Never mind there are no reported cases of wrongful foreclosure, the federal government has sued virtually every large bank, extracted “settlements” typically in excess of $1 billion, and required them to enter into agreements more onerous than Dodd-Frank. The banks that entered into settlements and consent decrees include the following: Wells Fargo, Bank of America, CitiBank, PNC Bank, EverBank, JPMorgan Chase, One West, Santander Bank, and U. S. National Bank. Loan servicer Ocwen entered into a settlement with the attorney generals in 49 states and with the federal government.

From the writer’s perspective, the volume of loan defaults escalated at a rate no one expected, and the loan servicers did their best to handle the defaults. The lawsuits by the federal government and attorneys general were a stalling action to delay the foreclosure activity action and to extract money from legitimate businesses with little reason.

This article breaks loan servicing into four simple segments: 1) on-boarding, 2) routine serving, 3) preliminary default, and 4) default requiring foreclosure or a workout. Each segment is examined along with thoughts on how  Big Data can be integrated.


Many loan servicers handle either paper documents or scanned documents that cannot be read or queried using PDF or an optical character recognition file. The first critical stage is to generate loan documents with a loan processing system that generates images as well as indexes to the images. First, this will sharply reduce the time to on-board a file. Second, it will make it possible to do validations not possible without an electronic file. Consider validation of the following issues:

  • Is this the borrower’s primary residence, or is he also claiming multiple other first residences?
  • Is the reported sales price consistent with the assessed value for property taxes?
  • Verification of borrower identity using electronic contact with third party service.
  • Extracting key loan and deed of trust terms to affirm they are consistent with the tape or other documentation presented.
  • Confirming that all required documents, based on state law and location, have been received. For example, in Texas, a municipal utility district (MUD) document is required if the property is in a MUD. In certain coastal areas, borrowers must affirm they understand the risk of storms.
  • Electronic loan processing systems can provide a list of completed documents and confirm whether all documents have been completed.
  • Are the dates in the documents consistent with the loan?
  • Are there additional documents, such as a release of lien, that have been filed?
  • Is the current flood determination and flood policy appropriate?
  • Is the monthly property tax escrow reasonable based on the assessed value and tax rates? How about if the property is new construction purchased in December which will be reassessed on full value in January?
  • If the loan is being on-boarded from another servicer, have all appropriate documents been produced?

All of these tasks can be done manually, but often would not be. However, all of these tasks can be performed electronically with the proper system.

Routine Servicing

During routine servicing, a  Big Data platform can minimize costs, maximize collections, and likely minimize regulatory trouble. We will next review events other than monetary defaults and then monetary defaults. Lenders will seldom pursue foreclosure for a non-monetary default. However, they should be aware of and be tracking such events.

Non-Monetary Defaults

Non-monetary defaults should be considered an early warning system of possible problems. Following are examples of warning signs:

  • A borrower has routinely made payments five to 10 days prior to the first of the month. However, they now are arriving late.
  • A borrower has routinely been making monthly payments of $500 in addition to the principal, interest, taxes and insurance (PITI); however, they revert to the basic minimum payment.
  • The borrower is sued by the home owners associates (HOA) for not paying HOA dues.
  • An abstract of judgment is filed against a debtor.
  • The borrower’s mailing address, which has been the collateral for the loan, changes to another city or state.
  • The local government files a lien against the collateral claiming the property to be uninhabitable.
  • The borrowers, a married couple, file for divorce.

While each of these in the abstract seems minor, they do contain information. When evaluated over a portfolio of one million or more mortgages, the implications of each can be evaluated.

Monetary Defaults

Reports have indicated that few loan workouts have been successful. At the very least,  Big Data can help to determine which are likely to be successful and which are not. However, early contact with the borrower to understand the circumstances and to evaluate their chance of success can be both humane and financially appropriate.  Consider the following events of financial default:

  • George is a gardener. He just bought a $900,000 home using a stated-income loan. He missed the first payment. The stated income was $300,000. When you discuss George’s income and how much he can afford for a monthly payment, he indicates between $500 and $600 per month. The best option is likely to make a deal with George to get him to move, perhaps giving him moving money, and proceeding with foreclosure.
  • Ron is a skilled machinist, married for 30 years. He bought a $200,000 home five years ago and is 15 days late on his payment for the first time. He reports his prior employer closed without notice and he has three offers at a similar or higher wage rate. Assuming the servicer believes Ron, it would make sense to extend him every courtesy, work closely (weekly?) with him to get a payment plan that works and get him back on schedule.
  • Robert is a skilled geologist in west Houston who lives in a $700,000 house. He bought the house 12 years ago, and this is the first time he has missed a payment. The mortgage is now $400,000. Robert was earning about $250,000 annually. His wife earns $30,000 as a teacher. You call him 15 days after he misses the payment, optimistic that this is clearly a borrower who can get back on his feet. Unfortunately, he reports he worked for a small oil company and does not have a pension plan and has literally no savings. In addition, he has mailed 1,200 resumes attempting to find work and has not yet had one interview. A reasonable approach is to encourage him to immediately start to market his house before payments and late fees erode his equity.

The objective with the monetary defaults has not been to develop an exhaustive list of possible options. It has been to illustrate that monetary defaults vary and can be categorized. In addition to the monetary default information, all other information about the borrower, loan, collateral, metro area, and payment history can be categorized. By studying and categorizing historical results considering many factors, evaluating the attempts to resolve them and the success rate of various options, you can reduce loan losses and work with borrowers more realistically. The ability to do this analysis on the fly greatly exceeds the capacity of even the brightest person.

Some of the monetary default cases will best be resolved by foreclosure. Even the permutations of handling the foreclosure can and should be studied. Some borrowers will fight you every step of the way. Some borrowers will be eager to retain their credit and be more reasonable. The borrower’s history and the data you have chronicled on their monetary and non-monetary defaults will help you to understand which option makes sense.

Given the federal and state government’s harsh response to foreclosing on delinquent loans, it makes sense to review options to maximize proceeds while antagonizing the borrower as little as possible. For securitized loans, there may be limits on the creativity available to quickly and cost effectively obtain possession of the collateral without motivating the borrower to trash the property.

Integrating Documents, Data, and Decision Making

Compiling the databases described and preparing analytical and graphical tools will indeed require technicians with special expertise. However, the goal is to provide distilled data to people making decisions. In the past, a customer service representative at a loan servicer may not have had authority to allow a short sale or restructured loan. By delivering the right information to the right people, they are enabled to make the right decision. In some cases it will still be necessary to obtain information from the borrower, or to refer the decision to a higher level. In addition to data available from loan documents and the payment history, many sources of data are available to enrich the quality of the data and empower the decision makers.

Probably the most exciting aspect of  Big Data is expediting distilled information which has been analyzed to people who can make decisions. In some cases these will be C-level executives. However, in progressive organizations, decision making will be allowed at the lowest possible level, by those who are closest to the facts and information.

Patrick O’Connor is president of Enriched Data, which maintains a national database for residential and commercial real estate. He can be reached at PatrickO@POConnor.com.

Be Sociable, Share!
(Visited 231 times, 1 visits today)

Check Also

Avoiding Surprises

Servicing Quality Assurance and Quality Control Monitoring Systems can be Management’s Best Friend By Jim …